Model · alpaca-lora-7b
alpaca-lora-7b
Developer: tloen (Alpaca-LoRA project; individual developer, name not_recorded on the sources read)
Availability: available · checked 2026-09-24 · source
Raw record: /data/models/alpaca-lora-7b.json
Fields
- id
- alpaca-lora-7b
- identifiers
- huggingface
- tloen/alpaca-lora-7b
- developer
- tloen (Alpaca-LoRA project; individual developer, name not_recorded on the sources read)
- release_date
- 2023-03-14partial · sourcenote: HF repo creation date; may predate or postdate public release.
- weights_status
- open
- availability
- available · checked 2026-09-24 · source
- license
- mitpartial · sourcenote: From HF card metadata (`license: mit`); covers the adapter only. The card does not address the LLaMA base license. GitHub LICENSE not checked.
- architecture
- family
- decoder_only
- n_layers
- 32recorded · source
- hidden_size
- 4096recorded · source
- n_heads
- 32recorded · source
- vocab_size
- 32000recorded · source
- positional_encoding
- rotary (RoPE)recorded · source
- context_length
- 2048recorded · source
- note
- Propagated from llama-7b: the card says "a low-rank adapter for LLaMA-7b" and adapter_config.json names a LLaMA-7B checkpoint (decapoda-research/llama-7b-hf, a third-party re-upload) as base. The adapter changes no architecture field.
- training_data
- Stanford Alpaca instruction data, per the card's metadata as the cleaned version yahma/alpaca-cleaned. Card prose: 'fit on the Stanford Alpaca dataset'.partial · sourcenote: Card metadata lists yahma/alpaca-cleaned and the project README's training command uses --data_path 'yahma/alpaca-cleaned'; the card prose links the original Stanford Alpaca repo. The two differ (cleaned vs original 52K); the card does not settle which this checkpoint saw. The HF repo dates from 2023-03-14, possibly before the cleaned set existed.
- techniques
- primary_sources
- record_history
- date:2026-09-24 · change:ingested as candidate from HF (tloen/alpaca-lora-7b@12103d6baae1b320aa60631b38acb6ea094a0539) · by:ingest_hf.py ·date:2026-09-24 · change:preparer: developer, architecture (propagated from llama-7b, adapter-only), training data (partial), 2 edges (adapter_on, trained_on); sources: card, adapter_config.json, alpaca-lora README · by:claude (preparer, Sonnet 5) ·date:2026-09-24 · change:reviewed and promoted from staging (2 edge(s) accepted) · by:Wilson Pruitt ·
Parents
Weights descend
- adapter_on → LLaMA 7B declared source Card: 'This repo contains a low-rank adapter for LLaMA-7b fit on the [Stanford Alpaca] dataset.' adapter_config.json: peft_type LORA, r 16, targets q_proj,k_proj,v_proj,o_proj; base_model_name_or_path decapoda-research/llama-7b-hf (a third-party re-upload of LLaMA-7B; the card names LLaMA-7b). The record holds adapter weights only; a runnable model needs LLaMA-7B.
Training data
- trained_on → Alpaca-Cleaned declared source Card metadata: datasets: [yahma/alpaca-cleaned]; README training command --data_path 'yahma/alpaca-cleaned'. Uploader tloen is the developer. Card prose instead says 'Stanford Alpaca dataset'; see training_data note. Reviewer: if you prefer the prose reading, swap to alpaca-52k.
Children
No edges recorded.
Read in
No station on the reading path has touched this record yet.